Japan’s Trying Hard to Catch U.S. in Software : Programming: The U.S. has 70% of the world market. Some say this proves that the Japanese aren’t creative, but Americans in the industry are nervous.

In February, Nikkei Computer magazine rendered its verdict on one of the largest software development projects attempted in Japan, an automation project known as SIGMA.

“The Failure of a National Project that Took Five Years and 25 Billion Yen,” the headline blared.

Software specialists have come to a similar conclusion about the Fifth Generation Project, the multimillion-dollar endeavor that was supposed to propel Japaninto world leadership in artificial intelligence.

“The stated goal was to leapfrog generations, and from that point of view, it was a flop,” said Edmond Schonberg, a New York University professor who surveyed Japan’s software development for the U.S. Commerce Department in 1986.

In a seemingly relentless march toward technological supremacy, Japan has dazzled the world with printers and copiers, autos and VCRs, stereos, semiconductors and supercomputers. But despite project after project, millions of investment dollars and national prestige being on the line, the Japanese are struggling with what many regard as the future’s key technological frontier: software.

“I believe the one who can control the software can control business,” said Katsuhide Hirai, director of Fujitsu America Inc.’s information systems division.

Yet it is Japanese stereos that the world demands, not their records; their VCRs, not their movies. In the same way, software specialists say, the Japanese have been far stronger in selling their business machines than the written instructions that tell them how to process words, calculate equations, analyze molecules, draw a three-dimensional architectural design.

In this arena, Made in America still dominates. U.S. software producers hold about 70% of the $70-billion world market, which is projected to explode to $1 trillion by the year 2000. To some analysts, the consistent inability of the Japanese to crack that market is proof that they lack the creativity and imagination critical to a process many liken to art.

“You can’t pick it up and imitate it so easily. You can’t carve it, measure it; you can’t even see it. You need more creativity, and we may have the edge there for a long time,” said Doug Jerger of ADAPSO, a software industry group.

“Software is pure mind stuff, more similar to poetry than anything else,” said Joe Garber, principal at A. T. Kearney Technology in Redwood City, Calif. “And people with artistic natures do tend to behave a little differently.

“We tend to be very tolerant of strange people with long hair and bad eating habits who produce things of genius. And the Japanese don’t quite understand that. The Japanese salaryman (office-worker) syndrome is not one that fosters artistic, aesthetically oriented creativity,” Garber added.

Sound like the smug complacency that undermined U.S. leads in autos, steel, semiconductors? It isn’t. Jerger and Garber, like nearly every American who discussed the issue, are not taking U.S. dominance for granted.

They recognize that the Japanese are not pursuing creative software with the same vigor that they are directing at breakthroughs in biotechnology, say, or new materials. They understand that the Japanese, pressed by a shortage of software engineers and buried by a backlog of orders, are pouring most of their energy into solid, reliable--but relatively unimaginative--projects, such as productivity tools and basic industrial systems.

Still, even as they rate Japan up to five years behind in most software areas, U.S. analysts throw in several big, cautionary buts :

* But the Japanese government has made software development a top priority. The Ministry of International Trade and Industry has promoted public-private programs and arranged tax incentives and software-engineer training programs.

* But major Japanese firms are pouring millions of dollars into the field. For instance, the 150% boost since 1981 in Hitachi Ltd.’s annual research spending of $2.7 billion has been driven by software spending in both basic industrial systems and such exotic-sounding fields as fuzzy logic. That form of software programs machines to think flexibly, like a human.

“Usually Hitachi focuses on manufacturing hardware,” said chief engineer Sumihisa Kotani. “But as many people say, how to combine these components to build an integrated system is very, very important. Top managers have decided that software is the key.”

* But the Japanese have proven their talent by building superior software systems that few Americans ever see: for banking operations, steel manufacturing controls, automobile plants. For instance, the auto industry’s processes of quality control and flexible manufacturing are largely driven by skillful software engineering. “That is the reason why Japanese auto manufacturers are beating America,” said Kouichi Kishida, technical director of Japan’s Software Research Assn. “The main weapon is the computer system.”

* But the Japanese have pulled even in key fields, such as supercomputer software, and are ahead in others, such as fuzzy logic. Rather than the yes-no, black-white thinking of most software programs, fuzzy logic allows sort ofs and shades of gray. The result: Japanese subway systems that give smoother rides, air conditioners that control humidity.

* But Japanese software tends to be more reliable. A 1984 study estimated that its error rate was only 10% of American software’s. When U.S. contractors for the Strategic Defense Initiative wanted to check their artificial intelligence programs, they turned to Fujitsu.

“Our engineers checked the program and corrected a lot of little things, step by step,” Hirai said.

* But the Japanese are advancing in software automation and boosting productivity by reusing more software code from program to program. More automation will free up programmers for creative design.

And what Japanese firms lack, they’re finding abroad. NEC Corp. is contracting for software on mainland China. Hitachi and Fujitsu have hired Americans for their U.S. software operations. Two years ago, Nomura Research Institute, a unit of the Japanese securities firm Nomura, commissioned a study of 10,000 U.S. software firms for investment and importing potential. Only two deals have occurred so far, but Garber said interest is picking up.

Why the scramble for software? “Because software is, “ ventures Jerger, who counts himself among the breed of bohemian software designers occasionally known to wear swami outfits and place pyramids atop their computers for inspiration.

Japanese executives, however, supply more dollars-and-cents answers. As more players sell hardware, dropping the price, software has become the critical factor adding value and distinction to a product. Software accounts for more than half a computer system’s cost today, Hirai figured, compared to about 20% five to seven years ago.

And, increasingly, software is what makes or breaks multimillion-dollar deals. Say a customer wants software to simulate auto crashes using a supercomputer and 100 workstations, a total $30-million deal. “If you cannot provide a crash analysis program, you lose $30 million,” Hirai said.

In the same vein, Sony Corp. learned the importance of software the hard way. After producing the world’s first commercial videocassette player, the Betamax, it virtually lost the market to the competing VHS standard. Sony’s investments in Columbia Pictures and CBS Records are aimed in part at controlling its software destiny.

Despite such huge stakes, Garber said most U.S. software firms are ignorant of Japan’s competitive potential. So key organizations are trying to alert them.

Last year, the ADAPSO software industry association published a report warning of Japanese strides in the field. Americans still lead, but given the competitive turnabout in semiconductors and other fields, “Japan’s extraordinary commitment to software must be a matter of concern,” the report said.

“I’m unwilling to be dismissive of them. Most of our clients have had that attitude, and they’re bleeding and bruised,” said the author of the ADAPSO report, Thomas R. Howell.

U.S. researchers are also monitoring Japanese progress. In 1984, researchers with the U.S. Commerce Department’s Japan Technology Evaluation Program concluded that the Japanese ranked even in product engineering, behind in advanced development and far behind in basic research. An upcoming follow-up concludes they are still generally behind. But Michael Harrison, a computer science professor at the University of California at Berkeley, who led the JTECH group last year, said they’ve made “enormous progress” in graphics and supercomputers.

Japanese companies are pinning their hopes on people like Shigeru Sasaki to help them close the software gap. The Fujitsu Laboratories Ltd. engineer helped develop a color image-processing system that sees as people do. The system has been used in experimental driverless cars, which follow white lines and recognize obstacles, and in surveillance systems.

In 1988, Fujitsu singled Sasaki out--among 3,000 researchers--as a person whose talents should be further developed. He spent a year at Carnegie Mellon University studying computer vision--and U.S. software development.

“Japanese companies are for product research,” Sasaki said. “If we want to study pure basic research, we need to go to the United States. U.S. universities have a lot of money and a lot of freedom and a lot of time to think deeply in the research field.”

Sasaki concluded that one of Japan’s problems is lack of trained specialists. In Japan, the average software professional is a high school graduate trained by his company. In the United States, a university-trained computer science graduate is more common.

Relatively few Japanese researchers pursue doctorate degrees, Sasaki said, because they wouldn’t get more money on the job--only a six-year delay in starting their career.

Japanese officials cite other reasons for the country lagging in software. The most prominent is that Japan has never developed a robust, independent software industry like the United States. That’s because more than 80% of Japan’s software is custom packages, specific programs written for a specific customer--a banking system, for instance. Typically, computer makers have supplied the programs along with the hardware.

In the United States, most software is “packaged”--that is, generic programs written by independent programmers. Much of the time, it is sold off-the-shelf in places such as bookstores. The Lotus 1-2-3 spreadsheet is an example.

Japan’s approach is considered inefficient. It takes only a few people to write packaged software for use by millions, but custom packages can tie up dozens of programmers on orders for just one customer. This exacerbates Japan’s most critical problem--a 250,000-programmer shortage that is projected to rise to 965,000 by 2000. It is also blamed for tying up talent that could otherwise be writing creative programs.

Japan’s “software factories” exemplify the different approach. In them, hundreds of programmers churn out lines of software code according to preordained rules. The group endeavor has its strengths: 70% more lines of code per person with half the errors compared to U.S. programmers, according to the upcoming JTECH report.

“They don’t have the wizards we do,” Harrison said. “But you do find a lot of disciplined people working together as a team to produce good, reliable software. Not brilliant, but solid and dependable.”

Which is not to be sneered at. Some industry experts, such as former TRW executive Barry Boehm, said the United States needs more of that reliability. He has written a well-known software model, Project Risk, for promoting it.

“If you are pure inspiration, then you maybe catch somebody’s imagination, but when they try to use that product day to day, they may get a little frustrated because it has lots of bugs in it,” said Boehm, a UCLA adjunct professor and director of the information science and technology office for the Defense Advanced Research Project Agency.

“On the other hand, when it becomes too rigorous . . . you miss the opportunities for what new dimensions the computer can bring to information processing. Japan tends to come down on the sort of factory approach to building software, and in the United States the tendency is to come down on the art-studio side. You need the right mix of both.”

Top Japanese firms say they prefer to focus on big-ticket industrial software packages because they are less risky and more profitable. A person who buys a word processing package may never give a company repeat business, but a utility or bank is likely to order another million-dollar program if satisfied with the firm’s quality and service, Hirai said.

In Japan, independent software designers have also been discouraged by a fractured market. While U.S. designers can find a huge market writing for one or two standards--International Business Machines, Corp. or Apple Corp.--the Japanese have had to consider not only IBM, but also Fujitsu, Hitachi, NEC and others. That’s like a recording artist having to compose separate musical scores for each different stereo model.

The small personal computer market is another disincentive. The Ministry of International Trade and Industry reports that Japan uses only a fourth as many PCs as the United States--5 million versus 20 million--even though the population is half as large. Executives speak of a national “keyboard allergy,” an inexperience-based aversion to computers.

That problem highlights another. The Japanese language, comprised of thousands of ideographic characters and two sets of phonetic alphabets, is far more difficult to use in software than the 26-character English alphabet. For starters, each kanji, or Chinese ideograph, takes twice as much computer space as a letter.

Officials also lament a lack of strong university programs in computer science--facilities where would-be programmers can play--and a culture that discourages youthful enterprise.

“Our social culture doesn’t seem to like creative young people. I think it’s the Confucian philosophy of just following your parents and teachers,” Kishida said. “I think the potential of Japanese programmers is almost the same or better, but there are comparatively small numbers of opportunities for these young, bright guys.”

In addition, Japan’s military establishment does not fund a lot of new software research, as the U.S. military does, analysts say.

And, in contrast with past successes in computer memory chips, Japan’s government “industrial policy” programs in software have been failures. Executives say central government planning worked well when Japanese companies were weak and new industries still small, but that world has changed.

“Unfortunately, MITI (the Ministry of Trade and Industry) doesn’t understand the computer industry. The technology is changing so rapidly that anything they plan and do becomes obsolete overnight,” said Tom Kobayashi, research director for Digital Equipment Corp. in Tokyo.

For Americans, the bottom line is that Japanese software producers do not pose a competitive threat--for now.

Commercially, very little Japanese software is exported to the United States. The Foster City, Calif., subsidiary of Dynaware Japan is nibbling at some markets, with products such as its 3-D architectural design package. Hitachi entered the U.S. market in 1987, but its graphics package failed to take off, so it has refocused on the mapping market. ASCII Corp., Japan’s leading software publisher, opened a U.S office in April but plans to export American packages to Japan rather than the reverse, officials said.

In contrast, U.S. firms have done better in Japan. Microsoft Corp. owns a 65% share of Japan’s market for personal computer operating systems, the set of instructions that tell computers how to run specific programs. Lotus Development Corp.’s 1-2-3 program is Japan’s best-selling spreadsheet package. More than 300,000 have been sold.

Besides, officials say, Japan is too busy with its own needs. MITI estimates that software demand is growing by 26% annually, but the number of software engineers is rising by only 13%.

“Japanese companies don’t have to be going abroad,” said Akihiro Sawa, a MITI deputy director. “There is enough market in Japan. Too much market.”